Case Study: Intelligent Lot Tracking

Modern semiconductor chip factories are among the most advanced production facilities in existence today. The cost of building a next-generation chip factory can easily exceed $1 billion. 24/7 production involving hundreds of employees in large clean-room facilities yields billions of chips per year. The semiconductor production process is extremely complex. Typically, multiple circuits are grouped on a single wafer, which undergoes complex chemical processes in a clean-room environment to build up the circuits in layers. Several wafers are then placed in a wafer carrier for processing. These wafer carriers hold the wafers between the processing stages. Advanced wafer carriers such as a state-of-the-art FOUP (Front Opening Unified Pod) can be used to automatically unload the wafers in the next production bay and reload them again after processing. Up to 500 production steps per wafer may be required, using hundreds of different machines. Some of the larger semiconductor factories have full-scale material handling systems that automatically move wafer carriers between production bays. Many factories which require more flexibility and support for a broad product mix, wafer carriers still need to be transported between production bays manually. The figure below shows an example of an FOUP wafer carrier (sometimes called a “lot box”) and its path through a factory, also known as a “wafer fab”.

In wafer factories that don’t have an automated material handling system, the production process is generally managed on the basis of dispatch lists. The dispatch list defines the order in which the production lots have to be processed. One of the main problems here relates to the localization of individual wafer carriers, as many factories use thousands of these carriers. Manual processing is costly and prone to error.

This is where intelligent lot-handling solutions come in. These use indoor localization technology to automatically track the position of each wafer carrier in a production facility. Positioning data is managed in a central database, which is closely integrated with the Manufacturing Execution System (MES).

Signaling devices, such as LEDs or markers (which changes colors from black to yellow) indicate whether a wafer carrier is currently scheduled for further processing. A display panel on the wafer carrier shows additional processing information, such as the lot number, next production operation or next destination of a lot box. However, implementing this kind of solution presents multiple challenges, as we will see below.

A good example of an intelligent lot-tracking solution is LotTrack®, developed by Swiss company Intellion, which we will examine in more detail in this case study. LotTrack is a system designed to improve the overall workflow in manually operated wafer fabs. It consists of three key components:

• DisTag: A smart device placed on each wafer carrier, which enables wafer carriers to be located within the factory to an accuracy of approximately 0.5 meters. The DisTag also has a control panel for local interaction with the factory operators. Signaling devices like an LED and a marker provide priority and search functions. Battery lifetime is approximately two years.
• Antenna Lines: The modular antenna line contains all hardware modules required for indoor localization, assistance and load port compliance. It is usually mounted to the ceiling of the clean room along the factory’s interbay and intrabay.
• Control Suite: The backend software is the link between shop-floor activities and the MES. It provides a dashboard for visualizing all transport and storage activities.

Customers like Infineon, STMicroelectronics or OSRAM use the LotTrack solution to reduce cycle times and work-in-progress (WIP), increase operator efficiency, digitize and automate paper-based administration processes  and enable automatic authentication of production lots at the equipment.

Technical Architecture

The Infineon plant in Villach, Austria is the headquarters of Infineon’s Automotive and Industrial Business Group, which mainly develops integrated circuits (ICs) for use in cars, such as engine control ICs. Flexibility is important for this factory, which produces approximately 800 different products with a total volume of 10 billion chips per year [LT1]. Because of the high number of different products and associated production process variations, the factory uses a manual transportation process for wafer carriers. Over 1,000 wafer carriers have to be managed simultaneously. In storage areas, over 16 wafer carriers can be stored per square meter. The clean room contains numerous elements that can cause electromagnetic reflection, such as the walls, production equipment, and storage racks.

These factors all make this kind of factory a highly challenging environment for a tracking solution. In particular, finding a technical solution for indoor localization that combines an acceptable cost factor with sufficiently high resolution is still a challenge (see section on indoor localization systems in the “Technology Profiles” section.) To address this problem, the LotTrack solution uses active and passive RFID (Radio-Frequency Identification) in combination with ultrasound technology. The antennas on the ceiling contain ultrasound emitters that periodically send out a ping signal. These ping signals are received by the DisTags on the wafer carriers. The DisTags compute the outward travel time of the ultrasound waves and temporarily store the results locally, together with the signal strengths. Using RFID communication, the ping signal analysis data is communicated back to the RFID receivers in the Antenna Lines. From here, this data is sent back to the central server. In the backend, a complex algorithm derives the real-time position information from the UHF (Ultra-High Frequency) pings sent from the antennas to the DisTags [LT1].

The system in place at Infineon Austria now processes three billion UHF pings per day (!). From this, around 270 million positions are calculated, to an accuracy of approximately 30 centimeters. About 500,000 position updates are communicated to the client systems each day. The system operates in near-real time, with the result that position changes by wafer carriers are recognized by the backend system within 30 seconds [LT1].

Conclusions and Outlook

The following is a summary of the key lessons learned from Intellion:

• Wafer factories with a diverse product portfolio are particularly in need of solutions that are more flexible than fully automated conveyor belts. Intelligent lot handling can provide the required flexibility if delivered as a modular system.
• These types of environment have very strict requirements. Ensuring 100% availability calls for significant investment and a sound infrastructure design.
• High precision for indoor localization depends on a combination of technologies (in this case, ultrasound and RFID). This is especially feasible in wafer fabs due to fab setup (i.e. long floors with straight branches between them).
• The need for maximum efficiency in system management should not be underestimated.
• Customers require long-term support, which means that the solution design and roadmap must be capable of dealing with multiple system versions in the field. The challenge facing product management teams is to efficiently manage advances in new technology and product versions. Downward compatibility becomes a major concern.

We would like to thank Kai Millarg, Managing Partner at Intellion for his support in writing this case study.